查看文章

The study objective was to develop a risk model incorporating diagnostic information to adjust for case-mix severity during routine monitoring of outcomes for pediatric cardiac surgery.

Data from the Central Cardiac Audit Database for all pediatric cardiac surgery procedures performed in the United Kingdom between 2000 and 2010 were included: 70% for model development and 30% for validation. Units of analysis were 30-day episodes after the first surgical procedure. We used logistic regression for 30-day mortality. Risk factors considered included procedural information based on Central Cardiac Audit Database “specific procedures,” diagnostic information defined by 24 “primary” cardiac diagnoses and “univentricular” status, and other patient characteristics.

Of the 27,140 30-day episodes in the development set, 25,613 were survivals, 834 were deaths, and 693 were of …